Extended model predictive control software framework for real-time local management of complex energy systems

Aragón Cabrera, Gustavo Alejandro; Jarke, Matthias (Thesis advisor); Monti, Antonello (Thesis advisor)

Aachen : RWTH Aachen University (2021)
Dissertation / PhD Thesis

Dissertation, RWTH Aachen University, 2021


Currently, the primary source for electricity generation are fossil fuels, which increase the emissions of CO2 aggravating the climate change on our planet. Therefore, a more sustainable form of energy generation is required, which can be accomplished with high penetration of renewable energy resources in the energy generation. Nevertheless, this transition changes the centralized generation model of the grid to a decentralized one that incorporates DG at different levels. Due to the problems associated with the increase of DG into the distribution network, such as voltage fluctuations, reverse power flow in the feeders, among others, the distribution network has to shift into a high controllable system converting it into a smart grid. New sensor and actuator technologies and EMS enabled by rapid ICT development are essential to counteract the problems associated to high DG and allow their integration into the energy supply.This thesis deals with the challenges of MPC-based EMS and the increasing integration of new sensor and actuator technologies to enable a smart grid. There is no evidence in literature of a higher abstraction about MPC that overcomes its initial definition. Therefore, this thesis introduces the concept of extended MPC framework that includes ICT aspects of the integration of new sensor and actuator technologies, forecasting methods, data management, flexible optimization model definition and solvers integration, in consideration of user needs and deployment requirements in real systems. Thus, the extended MPC framework facilitates the implementation and deployment of MPC as local EMS in energy systems. In order to define and implement the extended MPC framework, this thesis analyzes MPC from two different points of view. The first one is given by the challenges of users working with MPC collected through context interviews. The second one is derived from requirements for the local deployment of MPC systems in real test sites. The conjunction of both set the basis for the definition and implementation of the extended MPC framework. It supports the integration of heterogeneous sensor and actuator technologies, includes data management, forecasts implementation, a flexible optimization model management and the integration of various solvers.The extended MPC framework is validated through its deployment in three different use cases. The first use case includes discrete MPC systems embedding four different optimization models deployed locally in five residential buildings in Fur, Denmark. The second use case includes stochastic MPC systems embedding two different optimization models deployed in one residential building and one car park in Bolzano, Italy. Finally, the third use case links the extended MPC framework to a power flow simulator in order to allow simulations with ESS and charging stations for EV in a known grid model.


  • Fraunhofer Institute for Applied Information Technology [053300]
  • E.ON Energy Research Center [080052]
  • Department of Computer Science [120000]
  • Chair of Computer Science 5 (Information Systems and Databases) [121810]
  • Institute for Automation of Complex Power Systems [616310]